Pith. sign in

REVIEW 4 major objections 6 minor 28 references

Domain-adversarial Network Alignment

T0 review · 4 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that hiding which network a node belongs to, via an adversarial domain classifier, improves the matching of users across social networks.

desk verdict A clean, moderate-novelty combination of GCN embeddings, a DANN-style domain adversary, and a MAP anchor objective; the gains are plausible, but the experiments need error bars and a clear validation protocol before the numbers are fully reliable. read the letter →

arxiv 1908.05429 v1 pith:S5MW65ES submitted 2019-08-15 cs.LG stat.ML

Huiting Hong , Xin Li , Yuangang Pan , Ivor Tsang This is my paper · ORCID
classification cs.LGstat.ML
keywords networkalignmentdomainadaptationadversariallearninggraphconvolutionalnetworksanchorlinksrepresentationinvariancesocial
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that network alignment—matching the same user or entity across two different networks—is held back by domain-dependent features that embeddings learn, such as the average degree of each network. It proposes DANA, a unified deep architecture in which graph convolutional networks produce node embeddings while an adversarial domain classifier, equipped with a gradient reversal layer, pushes those embeddings to hide which network a node came from. The same objective maximizes a posterior probability over observed anchor pairs, treating alignment as a bi-directional matching problem. On three real-world social network datasets, the full model with weight-sharing and direction-aware convolutions (DANA-SD) reports Hits@1 of 22.97%, 18.42%, and 19.30%, above the best cited baselines. The paper argues this is because suppressing domain representation bias makes the learned representations more task-specific to alignment.

What carries the argument

The load-bearing mechanism is a minimax game between a domain classifier and the two graph convolutional feature extractors (Eq. 7). A gradient reversal layer (GRL) sits between the extractors and the domain classifier: in the forward pass it passes features unchanged, and in the backward pass it reverses the gradient sign, so one end-to-end update simultaneously maximizes the domain-classifier loss (making the representation domain-invariant) and minimizes the alignment loss. The alignment loss itself is a bi-directional posterior (Eq. 3): for each anchor pair, the log-probability of observing the match in both directions, where the conditional probabilities are sampled softmaxes over node embeddings (Eq. 4-5). This probabilistic view treats alignment as multi-view matching rather than as a distance-minimization or a projection-mapping problem.

What would settle it

Construct two synthetic networks with strongly differing degree distributions where the ground-truth anchor pairs are exactly those with matching degree rank (i.e., hubs link to hubs). If running DANA with a large domain-adversarial weight gamma produces lower Hits@1 than running DNA with the adversary removed, then the paper's central claim—that suppressing domain-dependent features always aids alignment—would be falsified for that regime. The same test could be run on the real datasets by artificially rewiring one network's degree distribution and tracking the DANA-vs-DNA gap.

Watch

Extended reading notes

Core claim

The central discovery the paper argues for is that injecting an adversarial domain classifier into the network-embedding process removes domain-dependent signals that are irrelevant—and sometimes harmful—to the alignment task, and that this removal measurably improves anchor prediction. Concretely, the paper trains two GCNs on the two networks, supervises them with a softmax-based posterior over observed anchors (Eq. 3), and simultaneously trains a domain classifier to tell which network a node came from while the GCNs are trained to fool it (Eq. 6-7, via gradient reversal). The ablation DNA, identical except for the adversarial module, consistently scores lower than DANA on all three datasets, which the paper takes as direct evidence that the adversary is what carries the gain. The paper also shows in a toy twinning-networks case that the adversarial training makes the domain classifier fail while diversifying the hidden-neuron directions, and interprets this as the disappearance of the y-axis-symmetric domain-discriminative pattern.

Load-bearing premise

The paper assumes that a representation which hides which network a node comes from still contains enough information to identify the same user across the two networks, even though structural signals like degree distribution can differ sharply and may be the very signals that disambiguate users.

Editorial extensions

If this is right

  • If the central claim is right, then any representation-learning approach to network alignment should explicitly suppress domain-dependent features rather than merely map one embedding space to another.
  • The reported superiority of DANA-SD over DNA, the no-adversary ablation, implies that the domain classifier is not a regularizer of convenience but the source of a large part of the improvement, so alignment methods built without it are leaving accuracy on the table.
  • Weight-sharing between the two GCNs helps alignment and speeds up training, suggesting that pushing the two networks into a common subspace is beneficial even before any matching layer is learned.
  • Direction-aware convolution (in-degree and out-degree views) improves link prediction within each directed network, which transfers to better alignment on directed social graphs like Twitter and Weibo.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable consequence the paper does not explore: the benefit of domain invariance should shrink or reverse when the domain-dependent signal is itself the best matching cue, for example when hubs in one network correspond to hubs in the other; one could vary the degree-distribution asymmetry between two synthetic networks and watch whether the DANA-vs-DNA gap flips sign.
  • The toy case only demonstrates that the domain classifier is fooled; it does not by itself prove the invariant representation retains enough discriminative power for alignment, so a formal information-theoretic statement of what is preserved under invariance would turn the empirical claim into a theorem about the method's limits.
  • The same adversarial-invariance recipe could transfer to other entity-resolution problems where domain bias is structural, such as cross-lingual knowledge-graph alignment or user linking across platforms with very different activity patterns, but the invariance target would need to be chosen per task since 'domain' is not always the network identity.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes DANA, a deep architecture for network alignment that combines GCN-based node embeddings with an adversarial domain classifier. The stated goal is to learn domain-invariant node representations by suppressing domain-dependent features that are irrelevant, or detrimental, to the alignment task. The training objective is a sum of a maximum-a-posteriori anchor-pair likelihood (Eq. 3) and an adversarial domain-classification loss (Eq. 6), optimized jointly through a gradient reversal layer (Eq. 7). Variants add weight-sharing between the two GCNs (DANA-S) and a direction-aware convolution structure for directed networks (DANA-SD). Experiments on DBLP, Foursquare-Twitter, and Douban-Weibo report Hits@1 and MRR improvements over several baselines, including a DNA variant that removes the domain-adversarial component. The paper also includes sensitivity analyses for the number of GCN layers L and the adversarial weight gamma, a comparison of the proposed probabilistic objective with an MSE objective, a link-prediction check of the directed convolution, and a toy case study of domain-invariant embeddings.

Significance. If the central empirical claim is supported, the paper makes a useful contribution by transferring domain-adversarial training from classification to network alignment, and by isolating the effect of the adversarial component through the DNA ablation. The MAP-based objective is derived cleanly, and the ablation ordering (DANA-SD > DANA-S > DANA > DNA) is logically organized and consistent with the paper's motivation. The paper also provides a direction-aware GCN variant that appears to help on directed networks. However, the empirical evidence would be much stronger with repeated-run statistics, a clearly separated validation protocol, and code or data release; the current point estimates without variance do not fully protect the main comparison against selection effects or run-to-run variability.

major comments (4)
  1. [Section 4.2 / Figs. 7 and 8] The sensitivity analyses for the number of GCN layers L and the adversarial weight gamma are evaluated with the same test-set metrics (Hits@k and MRR) that form the headline results in Table 2, and the paper does not state that a held-out validation split was used for model selection. With only single-run numbers in Table 2, the reported advantage of DANA-SD over DNA could be inflated by selection bias rather than caused by the adversarial module. Please specify the selection protocol, and ideally add a validation split or nested evaluation so that the chosen hyperparameters are not selected on the test set.
  2. [Section 3.1.1 Eq. (4) and Section 4.1.1] Training uses a dot-product softmax to define the anchor likelihood in Eq. (4), while evaluation ranks candidates by cosine similarity. The paper does not state that the learned representations are normalized, so the training and scoring objectives are not matched. A dot-product objective can be dominated by vector magnitude, whereas cosine similarity is scale-invariant, so this mismatch could materially affect the reported ranking results. Please clarify whether the representations are normalized, or justify and explicitly test the effect of this train-scoring mismatch.
  3. [Section 4.3 / Fig. 10] The toy case study demonstrates that DANA-S can fool the domain classifier on a symmetric twin network, but it does not establish the paper's underlying assumption that a representation which hides domain identity still preserves enough anchor-discriminative information on real data. The case study shows only that invariance is achievable, not that invariance is sufficient for alignment. Please provide a quantitative analysis on real data, for example by measuring both domain-classification accuracy and alignment performance as gamma varies, or by testing whether fully or partially invariant representations retain the information needed to rank true anchors.
  4. [Table 2 and Section 4.2.1] No error bars, standard deviations, or repeated-run results are reported anywhere in Section 4, and each number in Table 2 appears to be a single point estimate. The central claim that the adversarial component improves alignment rests on comparisons such as DANA-SD versus DNA (e.g., MRR gains of 5.70%, 27.86%, and 63.20% on the three datasets), but without variance estimates these gains cannot be assessed for statistical significance. Please report results over multiple independent runs, with means and standard deviations or confidence intervals, at least for the key DANA/DNA comparisons.
minor comments (6)
  1. [Section 1, first paragraph of contributions] The text says 'vanila GANA' where 'vanilla DANA' is intended; please fix the typo.
  2. [Section 4.1.1, formula for Hits@k] The displayed formula 'Hits @k = Hits A@k + Hits B@k |St est|× 2' is ambiguous; parentheses should be added so that the division and multiplication are unambiguous.
  3. [Section 3.1.1, Eqs. (3) and (4)] The notation p(v_j^B | v_i^A) abbreviates a conditional distribution that also depends on the model parameters, but this dependence is stated only in the surrounding text and not in the equations themselves; please make the notation consistent.
  4. [Section 4.2.3, Eq. (9)] The symbol C is used both for the number of negative samples in Eq. (9) and for the candidate set size in Eq. (5); please use distinct symbols to avoid confusion.
  5. [Section 4.2.1, discussion of DBLP] The dataset name DBLP is misspelled as 'DLBP' in the sentence about the undirected network; please correct it.
  6. [Section 2.2, last paragraph] In the sentence 'Meanwhile, the presentation learning is also task-driven', 'presentation' should be 'representation'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DANA's adversarial alignment objective is an empirical architecture trained on held-out anchors; the DANA-vs-DNA comparison is an ablation, not a fitted result renamed as a prediction.

full rationale

The central claim is that adding an adversarial domain classifier to a GCN-based anchor-posterior objective improves network alignment. This claim is supported by training on 80% of anchor labels and testing on the remaining 20%, so the reported Hits@1 and MRR values are not forced by construction. The comparison DANA versus DNA is an ablation: DNA is defined as DANA without the adversarial component, and the difference is an empirical outcome rather than an identity. The objective in Eq. (7) combines the anchor posterior Eq. (3) with the domain-classifier loss Eq. (6) through a learned weighting; no equation reduces the test metric to a fitted parameter. The paper's reliance on DANN as a motivating framework is an external citation of a published method, not a self-citation, and it does not import any uniqueness theorem that forces the architecture. The only notable self-citation is that author Xin Li is a coauthor of the IONE baseline [11], but IONE is used as an external comparison method with its own published code and benchmark data, not as a premise in the derivation of DANA. Consequently, the self-citation is not load-bearing. Hyperparameter sensitivity analyses in Figs. 7 and 8 are empirical explorations and, while the paper does not specify a validation split, this is an experimental-design concern rather than evidence that a prediction reduces to its input by definition. Overall, the derivation chain is self-contained and the empirical comparisons are not circular.

Assumptions & free parameters 6 free parameters · 6 assumptions · 0 invented entities

The method rests on standard GCN machinery and the DANN domain-invariance assumption, plus a task-specific sufficiency assumption that domain-invariant features retain alignment information. Hyperparameters are hand-set and some are selected on test data. No new physical or theoretical entities are introduced.

free parameters (6)
  • gamma (adversarial weighting factor) = 1.0 (hand-set; sensitivity in Fig.8 on test data)
    Balances alignment loss and domain classifier loss; chosen after observing test-set Hits@1, no validation split described.
  • lambda (regularization factor) = 0.01
    Regularizes GCN and domain classifier weights; chosen by hand without validation.
  • L (number of GCN layers) = 2
    Sensitivity analysis in Fig.7 on test datasets shows 2 layers best; selected using test metrics.
  • embedding dimension = 100 (DANA-SD uses 50 per direction, concatenated to 100)
    Set unanimously for all models; affects capacity and results.
  • batch sizes U and Z = U=512, Z=training set size
    Training hyperparameters; Z set to full training anchor set, no ablation or validation.
  • learning rate = 0.001
    Adam optimizer default-like choice; not tuned.
assumptions (6)
  • standard math GCN layer-wise propagation rule with normalized adjacency and ReLU activation
    Used in Eq. (1) without proof, following Kipf and Welling [8].
  • domain assumption DANN principle: a representation that maximizes domain classifier loss is domain-invariant
    Borrowed from [5]; invoked in Section 3.1.2 and Eq. (6)-(7).
  • ad hoc to paper Domain-invariant representations are sufficient for network alignment
    Central thesis; if domain-specific signals are necessary for disambiguation, the method fails. No theorem or controlled real-data study supports this.
  • domain assumption Topological consistency assumption: affiliated nodes have consistent connectivity structure across networks
    Stated in Introduction, paragraph 2; underlies all embedding-based alignment methods.
  • domain assumption Transductive use of test node structure and test nodes' domain labels during training
    Algorithm 1 samples vertex batches from the full V_A and V_B for the domain adversarial loss, including test nodes; this uses unlabeled test structure without explicit discussion.
  • standard math Sampled softmax approximates full softmax
    Used in Eq. (5) following [7]; the candidate set size is not specified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Domain-adversarial Network Alignment." pith.science (2026). https://pith.science/paper/S5MW65ES

@misc{pith2026190805429,
  author       = {Pith},
  title        = {Pith review of: Domain-adversarial Network Alignment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S5MW65ES}},
  note         = {Machine review of arXiv:1908.05429}
}
read the original abstract

Network alignment is a critical task to a wide variety of fields. Many existing works leverage on representation learning to accomplish this task without eliminating domain representation bias induced by domain-dependent features, which yield inferior alignment performance. This paper proposes a unified deep architecture (DANA) to obtain a domain-invariant representation for network alignment via an adversarial domain classifier. Specifically, we employ the graph convolutional networks to perform network embedding under the domain adversarial principle, given a small set of observed anchors. Then, the semi-supervised learning framework is optimized by maximizing a posterior probability distribution of observed anchors and the loss of a domain classifier simultaneously. We also develop a few variants of our model, such as, direction-aware network alignment, weight-sharing for directed networks and simplification of parameter space. Experiments on three real-world social network datasets demonstrate that our proposed approaches achieve state-of-the-art alignment results.

Figures

Figures reproduced from arXiv: 1908.05429 by the authors.

Figure 1
Figure 1. An SVM trained domain classi€cation on 2D rep [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. ‡e Vanilla Architecture of DANA 2.1 Embedding-based Network Alignment Among the various representation learning-based network align￾ment approaches, the main di‚erence lies in the way (1) What kind of network embedding approach is leveraged? (2) Whether the multiple networks are projected onto the same low-dimensional subspace? [18] proposed a shallow model MAH to align the network mani￾folds by modeling social grap… view at source ↗
Figure 3
Figure 3. Unfolded structure for directed networks [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Detailed performance comparison on real-world datasets. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 6
Figure 6. Figure 6: Hits@50 vs. Training ratio on Foursquare-Twitter. [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 8
Figure 8. Figure 8: Sensitivity analysis of parameter γ [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Objective analysis of alignment task. directions in GCNs is bene€cial to the representation learning of di￾rected networks, and in turn bene€cial to the alignment of directed networks. 4.3 Case Study: Domain-invariant Embedding To beŠer illustrate the characteristic of…
Figure 10
Figure 10. Figure 10: Hidden neuron visualization on the toy twinning [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 24 canonical work pages

  1. [1]

    Lgl: creating a map of protein function with an algorithm for visualizing very large biological networks

    Alex T Adai, Shailesh V Date, Shannon Wieland, and Edward M Marco/t_te. Lgl: creating a map of protein function with an algorithm for visualizing very large biological networks. Journal of molecular biology , 340(1):179–190, 2004

    work page 2004

  2. [2]

    Asnets: A benchmark dataset of aligned social networks for cross-platform user modeling

    Xuezhi Cao and Yong Yu. Asnets: A benchmark dataset of aligned social networks for cross-platform user modeling. In Proceedings of the 25th ACM International on Conference on Information and Knowledge Management , pages 1881–1884. ACM, 2016

    work page 2016

  3. [3]

    Adversarial network embedding

    /Q_uanyu Dai, Qiang Li, Jian Tang, and Dan Wang. Adversarial network embedding. In /T_hirty-Second AAAI Conference on Arti/f_icial Intelligence, 2018

    work page 2018

  4. [4]

    Convolutional neural networks on graphs with fast localized spectral /f_iltering

    Micha¨el Defferrard, Xavier Bresson, and Pierre Vandergheynst. Convolutional neural networks on graphs with fast localized spectral /f_iltering. InAdvances in neural information processing systems, pages 3844–3852, 2016

    work page 2016

  5. [5]

    Domain- adversarial training of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pascal Germain, Hugo Larochelle, Franc ¸ois Laviole/t_te, Mario Marchand, and Victor Lempitsky. Domain- adversarial training of neural networks. /T_he Journal of Machine Learning Research, 17(1):2096–2030, 2016

    work page 2016

  6. [6]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In Advances in neural information processing systems , pages 2672–2680, 2014

    2014

  7. [7]

    On using very large target vocabulary for neural machine translation

    S´ebastien Jean, Kyunghyun Cho, Roland Memisevic, and Yoshua Bengio. On using very large target vocabulary for neural machine translation. arXiv preprint arXiv:1412.2007, 2014

    arXiv 2007

  8. [8]

    Semi-supervised classi/f_ication with graph convolutional networks

    /T_homas N Kipf and Max Welling. Semi-supervised classi/f_ication with graph convolutional networks. arXiv preprint arXiv:1609.02907, 2016

    arXiv 2016

Show all 28 references
  1. [9]

    Deeper insights into graph convo- lutional networks for semi-supervised learning

    Qimai Li, Zhichao Han, and Xiao-Ming Wu. Deeper insights into graph convo- lutional networks for semi-supervised learning. In Proceedings of the /T_hirty-Second AAAI Conference on Arti/f_icial Intelligence, (AAAI-18), the 30th innovative Applications of Arti/f_icial Intelligen...

    2018

  2. [10]

    Adversarial learning for weakly-supervised social network alignment

    Chaozhuo Li, Yukun Wang, Senzhang Wang, Yun Liu, Philip Yu, Zhoujun Li, and Yanbo Liang. Adversarial learning for weakly-supervised social network alignment. In /T_hirty-/T_hird AAAI Conference on Arti/f_icial Intelligence, 2019

    2019

  3. [11]

    Aligning users across social networks using network embedding

    Li Liu, William K Cheung, Xin Li, and Lejian Liao. Aligning users across social networks using network embedding. In International Joint Conference on Arti/f_icial Intelligence, pages 1774–1780, 2016

    2016

  4. [12]

    Predict anchor links across social networks via an embedding approach

    Tong Man, Huawei Shen, Shenghua Liu, Xiaolong Jin, and Xueqi Cheng. Predict anchor links across social networks via an embedding approach. In International Joint Conference on Arti/f_icial Intelligence, volume 16, pages 1823–1829, 2016

    2016

  5. [13]

    User identity linkage by latent user space modelling

    Xin Mu, Feida Zhu, Ee-Peng Lim, Jing Xiao, Jianzong Wang, and Zhi-Hua Zhou. User identity linkage by latent user space modelling. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages 1775–1784. ACM, 2016

    2016

  6. [14]

    Cross-domain sentiment classi/f_ication via spectral feature alignment

    Sinno Jialin Pan, Xiaochuan Ni, Jian-Tao Sun, Qiang Yang, and Zheng Chen. Cross-domain sentiment classi/f_ication via spectral feature alignment. In/T_he 19th International World Wide Web Conference, pages 751–760. ACM, 2010

    2010

  7. [15]

    Deepwalk: Online learning of social representations

    Bryan Perozzi, Rami Al-Rfou, and Steven Skiena. Deepwalk: Online learning of social representations. In Acm Sigkdd International Conference on Knowledge Discovery & Data Mining , 2014

    2014

  8. [16]

    Evaluating web-based question answering systems

    Dragomir R Radev, Hong Qi, Harris Wu, and Weiguo Fan. Evaluating web-based question answering systems. In Proceedings of the /T_hird International Conference on Language Resources and Evaluation, LREC 2002, May 29-31, 2002, Las Palmas, Canary Islands, Spain, 2002

    2002

  9. [17]

    Modeling relational data with graph convolutional networks

    Michael Schlichtkrull, /T_homas N Kipf, Peter Bloem, Rianne Van Den Berg, Ivan Titov, and Max Welling. Modeling relational data with graph convolutional networks. In European Semantic Web Conference, pages 593–607. Springer, 2018

    2018

  10. [18]

    Mapping users across networks by manifold alignment on hypergraph

    Shulong Tan, Ziyu Guan, Deng Cai, Xuzhen Qin, Jiajun Bu, and Chun Chen. Mapping users across networks by manifold alignment on hypergraph. In Twenty- Eighth AAAI Conference on Arti/f_icial Intelligence, volume 14, pages 159–165, 2014

    2014

  11. [19]

    Arnetminer: extraction and mining of academic social networks

    Jie Tang, Jing Zhang, Limin Yao, Juanzi Li, Li Zhang, and Zhong Su. Arnetminer: extraction and mining of academic social networks. In Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining , pages 990–998. ACM, 2008

    2008

  12. [20]

    Line: Large-scale information network embedding

    Jian Tang, Meng /Q_u, Mingzhe Wang, Ming Zhang, Jun Yan, and Qiaozhu Mei. Line: Large-scale information network embedding. In Proceedings of the 24th inter- national conference on world wide web , pages 1067–1077, 2015. Domain-adversarial Network Alignment Conference’17, July ...

    2015

  13. [21]

    Irgan: A minimax game for unifying generative and discriminative information retrieval models

    Jun Wang, Lantao Yu, Weinan Zhang, Yu Gong, Yinghui Xu, Benyou Wang, Peng Zhang, and Dell Zhang. Irgan: A minimax game for unifying generative and discriminative information retrieval models. In Proceedings of the 40th International ACM SIGIR conference on Research and Develop...

    2017

  14. [22]

    Graphgan: Graph representation learning with generative adversarial nets

    Hongwei Wang, Jia Wang, Jialin Wang, Miao Zhao, Weinan Zhang, Fuzheng Zhang, Xing Xie, and Minyi Guo. Graphgan: Graph representation learning with generative adversarial nets. In/T_hirty-Second AAAI Conference on Arti/f_icial Intelligence, 2018

    2018

  15. [23]

    A survey of transfer learning

    Karl Weiss, Taghi M Khoshgo/f_taar, and DingDing Wang. A survey of transfer learning. Journal of Big Data , 3(1):9, 2016

    2016

  16. [24]

    Evaluation measures (information retrieval) — Wikipedia, the free encyclopedia

    Wikipedia contributors. Evaluation measures (information retrieval) — Wikipedia, the free encyclopedia. h/t_tps://en.wikipedia.org/w/index.php?title= Evaluation measures (information retrieval)&oldid=889157178, 2019. [Online; ac- cessed 21-May-2019]

    2019

  17. [25]

    Controllable invariance through adversarial feature learning

    Qizhe Xie, Zihang Dai, Yulun Du, Eduard Hovy, and Graham Neubig. Controllable invariance through adversarial feature learning. In Advances in Neural Information Processing Systems, pages 585–596, 2017

    2017

  18. [26]

    An information /f_low model for con/f_lict and /f_ission in small groups

    Wayne W Zachary. An information /f_low model for con/f_lict and /f_ission in small groups. Journal of anthropological research , 33(4):452–473, 1977

    1977

  19. [27]

    Integrated anchor and social link predictions across social networks

    Jiawei Zhang and S Yu Philip. Integrated anchor and social link predictions across social networks. In International Joint Conference on Arti/f_icial Intelligence, pages 2125–2132, 2015

    2015

  20. [28]

    Adversarial feature matching for text generation

    Yizhe Zhang, Zhe Gan, Kai Fan, Zhi Chen, Ricardo Henao, Dinghan Shen, and Lawrence Carin. Adversarial feature matching for text generation. arXiv preprint arXiv:1706.03850, 2017

    2017 arXiv

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.